From the Kitchen

00-blog-outline

Where AI Actually Belongs in Production Systems

Most AI demos put the model at the center. In production, the model is step 3 of 4. How we built a food matching pipeline where AI is the fallback, not the foundation.

LLMs Are Not Data Cleaners (Until You Force Them to Be)

We feed our LLM receipt photos, voice memos, and text blobs. It works — but only because we don't trust it. How regex-first cleaning makes LLM parsing reliable.

Why Embeddings Beat Filters (Then Don't)

Semantic search found recipes keyword search never could. Then it started recommending pasta for every soup query. How we built a hybrid search that balances four ranking signals.

AI Scoring Systems Drift (and How Relative Ranking Fixes It)

Our first AI scoring run rated 80% of recipes between 6 and 8 out of 10. The distribution was useless. Comparative batching with forced constraints fixed it.

Building a Recommendation Engine That Actually Explains Itself

Most recommendation engines are black boxes. Ours is a dot product you can read. 6 features, 3 lane profiles, and every score is interpretable.

What I Got Wrong Building an AI System (So Far)

42 database migrations, 2 AI agents, and a recommendation engine. Here's what I'd do differently — fuzzy alias poisoning, OCR hallucinations, embedding noise, and overengineering.